Interpolation raster for bedrock contours

[LukeRoyle]

Trying to create contours that represent bedrock across a site. There are 44 points in all, each having a base of bedrock record (AOD) at the site. When I've used some interpolations to try and create a raster for contouring from these point, but I am running into errors.

Mainly the interpolated bedrock surface appears to be above the surface at certain points on site, which does not match reality. I've got around this by creating a raster from depth below surface (i.g. 2m below surface) and deducting from surface, but this isn't accurate either, being generalised and incorrect at the points with a know depth.

Just wondering if anyone has experience with these issues and what I can do to create the most accurate surface estimation possible?

Thanks.

[Jon-PaulMcCool]

It will really matter what method of interpolation you’re using and the distribution of your data vs study area. For the study area, you need to avoid extrapolation as that is an easy way to get values outside the range of what is possible. For the interpolation method, have you tried EBK? I would avoid spline as it easily generates values outside the range of the possible. 
Another possible approach would be to create a TIN from the points. This way the surface matches the data perfectly at the observation points and would change linearly between them.